RACI matrix generator: assign roles and check them

How the RACI check works

A RACI matrix is only useful if it answers two questions for every task: who does it, and who answers for it. The generator therefore checks each row against two rules and shows a message in plain words for every row that breaks one. An unnamed row with no assignments is ignored.

One person can be both Accountable and Responsible, which is common in small teams. The generator has a combined A/R value for that case, and it counts for both rules.

Worked example

Take the task “Build and test the feature” in a launch with four roles. Engineering is R because it does the work, Product is A because it signs off, Support is C because it reviews the effect on customers, and Marketing is I because it just needs to know when it ships.

Read left to right the row is A, I, R, C. It has exactly one A and one R, so it passes the check.

Now mark Marketing as A too. The row has two Accountable roles and the generator reports that “Build and test the feature” has more than one. Change one back to I and the row passes again.

In the 5 tasks of the example, every row has exactly one A and at least one R, so the summary reads 5 of 5.

What R, A, C and I mean

The four letters describe a person’s part in a task, not their job title. Using a launch as the example:

• Responsible: the person or group that does the work. For the announcement that is the marketing writer. Several roles can share it.
• Accountable: the one role that answers for the result and has the final say, here the head of marketing. Only one, because shared ownership usually means no ownership.
• Consulted: people asked for input before the work is done, such as the product manager checking the feature description. Communication goes both ways.
• Informed: people told afterwards, such as support hearing the launch date. Communication goes one way.
• A blank cell means the role is not involved. That is useful information, not a gap to fill.

Common RACI mistakes

Most matrices that fail in practice fail in one of a few predictable ways:

• Two or no Accountable roles. This is the most frequent mistake and the reason the generator checks it.
• Too many Consulted. If everyone is consulted on everything, decisions slow down. Keep C for people whose input changes the outcome.
• Whole departments instead of roles. “Engineering” as Accountable does not tell anyone whom to ask. Name a role such as the engineering lead.
• Tasks that are too coarse or too fine. A row like “do the project” hides the hand-offs, and fifty rows hide the important ones.
• Making the matrix once and never revisiting it. Re-check it when the team, the process or the reporting lines change.

RACI vs RASCI, DACI and RAPID

RACI is the best known of several responsibility models and the simplest to read. RASCI adds an S for Supportive, people who help the Responsible role with the work without owning it. It is useful when assistance is a large share of the effort, at the price of a busier matrix.

DACI and RAPID are built for decisions rather than tasks. DACI names a Driver who runs the process, the Approver who decides, Contributors and Informed. RAPID adds Recommend and Perform. If your question is who decides, one of those may fit better; if it is who does what in a process, RACI is usually enough.

How to run a RACI session with your team

A matrix filled in by one person at a desk reflects that person’s view. A short session with the people involved takes about an hour and gives you a matrix they will actually use:

• List 10 to 20 tasks or decisions that matter and cause confusion. Skip routine work that is already clear.
• List the roles that take part, not the names of individuals.
• Fill in the Accountable column first, one per task, then the Responsible roles, and only then Consulted and Informed.
• Resolve every flagged row in the room. Disagreement about who is Accountable is the finding, not a nuisance.
• Copy the result as CSV into your wiki or spreadsheet and put a date on it.